FR2932045A1 - METHOD AND SYSTEM FOR PROTECTING THE DATA CONFIDENTIALITY OF A VIDEO STREAM DURING ITS TRANSMISSION - Google Patents

Abstract

A method of visually encoding at least a portion of a video stream or video sequence at least partially compressed, said stream being decomposable into a first type of objects and a second type of objects, the method being applying to each of the images contained in a video sequence characterized in that it comprises at least the following steps: analyzing the sequence in the compressed domain so as to define for a given image N at least a first group of objects to be protected by visual encryption and a second group of objects, (2, 3)> predicting from the data from the analysis in the preceding step of the compressed image N, the position of the objects for a next image N + 1, (4a)> determining the division into portions or groups of portions of the image N + 1, (4b)> compressing (8b) the first group of objects of the image N + 1 and encrypting at least a portion (8a),> transmit the other types of groups of objects for the image N + 1 to do step of compression (6).

Description

The invention relates to a method and a system for transmitting an at least partially compressed video stream while ensuring the security of the information to be transmitted. For this, the method includes a step of encrypting selected areas in a given image of the stream. The invention applies to transmit compressed video streams in a transmission context that can be monitored and to prevent access to the content of certain protected areas of the image. Thus, any malicious visualization of the flow can be avoided. It allows to protect the privacy of people for example in places equipped with video surveillance. In the remainder of the document, the term foreground designates the moving object (s) in a video sequence, for example, a pedestrian, a vehicle, a molecule in medical imaging. In contrast, the background designation is used with reference to the environment as well as to fixed objects. This includes, for example, soil, buildings, trees that are not perfectly immobile or parked cars. The invention can, inter alia, be applied in applications implementing the standard defined jointly by the ISO MPEG and the video coding group of the ITU-T called H.264 or MPEG-4 AVC (advanced video coding) which is a video standard providing more efficient compression than previous video standards while having a reasonable implementation complexity and directed to network applications. In the remainder of the description, the Applicant uses the expression stream being compressed or video sequence being compressed to designate the same object, namely: during compression means that the current image has not compressed when we execute the

steps leading to the encryption of the objects to be protected but that the previous images have already been compressed. The words encryption or encryption are used to designate the same transformation.

The expression slice or portion more known in the field under the Anglo-Saxon slice expression corresponds to a sub-part of the image consisting of macroblocks that belong to the same set defined by the user.

Currently, video surveillance is increasingly used, raising the issue of privacy. In some applications, only images or areas of video stream images must be accessible, accessibility being restricted to authorized persons.

The publication entitled Compiling Selective Encryption for H.264 / AVC video streams by Cyril Bergeron and Catherine Lamy. Proceedings of the International Workshop on Multimedia Processing, MMSP'05, Shanghai, discloses a method for partially encrypting a video stream to secure the transmission of information contained in this stream while preserving compatibility with the H.264 standard.

One of the objects of the present invention is to provide a method for protecting the image content or certain areas of an image in order to preserve the confidentiality of the data. This is, in this implementation example, executed during the step of compressing a data stream.

The invention relates to a method for protecting at least part of a video stream or video sequence at least partially compressed against a violation of the information contained in said stream, said stream being able to be decomposed into several types of objects, the process applying on

each of the images contained in a video sequence characterized in that it comprises at least the following steps: a) analyzing the video sequence in the compressed domain in order to define for a given image N at least one or more groups of objects to be protected by encryption, designated first group of objects, a second group of objects then being associated with the rest of the image, b) predicting from the data resulting from the analysis in step a) of the compressed image N , the position of the objects for a next image N + 1, c) determining the division into portions or groups of portions of the image N + 1, d) compressing the parts of the image comprising the group or groups of portions of the image N + 1 related to the first group and encrypting at least a part of the first group or groups of objects of the image N + 1, e) compressing the other types of groups of objects for the image N + 1 f) adding to the groups of objects of the image N + 1, resulting from steps d) and e) information ident checking the groups of objects or unencrypted. The method can add to the flows from steps d) and e), information identifying groups of encrypted objects or not. For H.264, for example, for a given image or group of images, an undefined NAL-type network transport unit (described in H.264 as undefined NAL) will be added, which will carry a information indicating which portions (slices) have been encrypted. The step of predicting the position of an object comprises, for example, the following 25 steps: using on the one hand the results of the analysis in the compressed domain leading to identified blocks within an image and also the stream being compressed, - follow up from these blocks identified by matching and prediction using the history of the positions of the objects on the N previous images.

The matching step may be based on a Munkres algorithm, and that of tracking on a Kalman filter. The method implements for the encryption step a selection of bits allowing decoding with a standard decoder, for example, according to the aforementioned article authored by Cyril Bergeron and Catherine Lamy. Since the video sequence is produced, for example, via the MPEG-4 part 10 / H.264 standard, the method for defining the portions or slices groups uses, for example, the flexible macroblock ordering (FMO) technique (Flexible Macrocbloc Ordering). ) authorizing the definition of groups of portions or slices macroblock groups by macroblock. The method can associate either a group of portions or slices group by object to protect or moving object, or a group slices to all moving objects (a second then being assigned to the background). The "image update" of the slices group is accompanied, for example, by the transmission of a parameter PPS (Picture Parameters Set) which indicates to the decoder the new division of the image. The invention also relates to a system for visually encrypting at least part of a video stream or video sequence at least partially compressed against unauthorized viewing of the information contained in said stream, characterized in that it comprises at least: a video part for the transmission of a stream of images during compression, a video coder, a stream selection module to be encrypted or encrypted and a suitable encryption or encryption module); the steps of the method comprising the aforementioned characteristics and a transmission module of the stream at least partially encrypted or encrypted.

Other features and advantages of the device according to the invention will appear better on reading the description which follows of an exemplary embodiment given by way of illustration and not limited to the appended figures which represent:

FIGS. 1 to 4, the results obtained by an analysis in the compressed domain, FIG. 5, an exemplary method according to the invention applied to a video stream being compressed, FIG. 6 an example of a diagram for a video encoder adapted to implement the encryption method according to the invention.

In order to better understand the operation of the method according to the invention, the description includes a reminder on how to perform an analysis in the compressed domain, as described for example in the referenced US patent application 2006 188013. Figures 1, 2, 3 and 4 and also in the following two references: Leny, Nicholson, Loaners, "Motion estimation for real-time video analysis in the compressed domain", GRETSI, 2007.

Leny, Loaner, Nicholson, "SPIE Electronic Imaging," San Jose, 2008. In summary, some of the techniques used in the MPEG standards discussed in these papers consist of splitting video compression into two step. The first step is to compress a still image. The image is first divided into blocks of pixels (of 4x4 or 8x8 according to the MPEG-1/2/4 standards), which are then subjected to a transform allowing a passage in the frequency domain, then a quantization allows to approximate or suppress the high frequencies to which the eye is less sensitive. Finally, the quantified data are entropically encoded. For this purpose, it allows to suppress or attenuate the high frequencies less sensitive to the eye and thus reduce the amount of information. The second step is to reduce time redundancy. For this purpose, it makes it possible to predict an image from one or more other image (s) previously decoded (s) within the same sequence (motion prediction). For this, the process seeks

in these images references the block that best corresponds to the desired prediction. Only a vector (Vector Motion Estimation, also known as motion vector), corresponding to the movement of the block between the two images and a residual error to refine the visual rendition are retained. These vectors, however, do not necessarily correspond to a real movement of an object in the video sequence but can be similar to noise. Different steps are needed to use this information to identify mobile objects. The work described in the aforementioned Leny et al publication, Motion Estimation for Real-Time Video Analysis in the Compressed Domain, and the aforementioned US Patent Application have delineated five functions making the analysis in the possible compressed domain, these functions and the modules implementing them are represented in FIG. 1: 1) a low-resolution decoder (LRD - Low-Res Decoder) makes it possible to reconstruct the entirety of a sequence at the resolution of the block , suppressing on this scale the motion prediction; 2) a motion estimation vector generator (MEG Motion Estimation Generator) determines meanwhile vectors for all blocks that the encoder encoded in "Intra" mode (within Intra or predicted images); 3) a Low-Res Object Segmentation (LROS) module relies on an estimation of the background in the compressed domain thanks to the sequences reconstructed by the LRD and thus gives a first estimate of the objects mobile; 4) a motion-based object filter (OMF) uses the MEG output vectors to determine moving areas from the motion estimation; 5) finally a module allowing to establish a cooperative decision (CD-Cooperative Decision) starting from these two segmentations, takes into account

the specificities of each module according to the type of image analyzed (Intra or predicted). The main interest of the analysis in the compressed domain relates to computation times and memory requirements which are considerably reduced compared to conventional analysis tools. Based on the work done at the time of video compression, the analysis times are currently 10 to 20 times the real time (250 to 500 images processed per second) for 720x576 4: 2: 0 images. One of the drawbacks of the analysis in the compressed domain as described in the aforementioned documents is that the work is performed on the equivalent of low resolution images by manipulating blocks composed of groups of pixels. As a result, the image is analyzed with less precision than using the usual algorithms used in the uncompressed domain. In addition, objects that are too small in relation to block cutting can go undetected. The results obtained by the analysis in the compressed domain are illustrated in FIG. 2 which show the identification of zones containing moving objects. Figure 3 schematizes the extraction of specific data such as motion estimation vectors and Figure 4 low resolution confidence cards obtained corresponding to the contours of the image.

FIG. 5 schematizes an example of implementation of the method according to the invention during which zones of an image being compressed, that is to say during the compression step, will be encrypted or This encryption step will be executed during the compression step implemented in a video transmitter comprising at least one video encoder and a processing unit as shown schematically in FIG. image having more importance in the flow in the sense of confidentiality will be chosen to be protected against undesired uses. The example given for illustrative purposes comprises two zones,

but nothing prevents a user from applying the method to several areas which will each have a degree of confidentiality associated, so different encryption keys and adapted to the level of confidentiality. The method and the system according to the invention implement the analysis in the compressed domain coupled to a tracking of objects better known under the name Anglo-Saxon tracking. This includes linking the independently segmented objects on each image to obtain a single, single object tracked during the sequence. For example, without tracking objects, If an object is present and segmented on 25 images, then the process will have to process 25 objects independently, which can bring multiple problems of merging tracks or indexing for example. The analysis of video sequences in the compressed domain presents, for its part, the advantage of using part of the work done by the video coder by exploiting the information available in the compressed domain. The method according to the invention will consider the images one after the other. It will select areas in an image that must be protected by encryption and apply the process to these areas. The stream 1 being compressed within an encoder is transmitted to a first analysis step 2. This step is executed frame by frame, whereas, just like the method, it can be implemented only on certain images. which will, for example, be determined and selected upstream of the process (output frequency less than that of the stream before compression, selection of certain images according to their type, etc.). The first step implemented during the method has the particular function of extracting data representative of an image that has been compressed. For a compressed image, the method has at the end of the first step, for example, a sequence of masks comprising blobs (regions having received the same label) related to moving objects 3 or foreground in some cases. The mask aims to separate in a compressed image the areas corresponding to the moving objects of the

other areas that are almost immobile, or background. The mask used can be a binary mask. By convention in this case, it is assigned the number 1 to designate foreground objects and the number 0 to designate the background. Any other convention may be used without departing from the scope of the invention. Instead of masks, the method according to the invention can also consider the enclosing boxes of moving objects. The bounding box coordinates correspond to the moving objects and are calculated using the mask. These boxes can be defined by two extreme points or by a central point associated with the dimensions of the box. It is possible in this case to have a set of image coordinates or a set for the whole of a sequence with the trajectory information (date and entry point, described curve, date and exit point). This method known to those skilled in the art will not be explained.

The analysis in the compressed domain can be carried out using the method described in the aforementioned US patent application. However, any method for obtaining an output of the analysis step in the form of masks per image, may also be implemented for the analysis step in the compressed domain.

Following this first step, a tracking or tracking step will be implemented. This tracking step can be performed from these blobs identified by matching, for example by applying the Munkres algorithm, then by position prediction, for example by Kalman filter, using the history of the position of the objects on the N previous images from the data resulting from the analysis of the compressed image N, as well as the results obtained equivalently on the previous x images; N û x to N. It is then possible to predict the position of the foreground objects in the next image 4a. This next image N + 1 has not yet been compressed, so it is still possible to determine the division into groups of slices, better known as

anglo-saxon slice group, adapted, which corresponds to step 4b. The analysis followed by the tracking step made it possible to predict for the image to be compressed zones containing mobile objects and others belonging to the background. The method will then define two types of groups of portions 5, 7 for the image. The first 5 will correspond to the background areas, which do not require visual cryptography. The second 7 will be dedicated to mobile objects, which the encryption or encryption step will target. There are two alternatives: declare all mobile objects in a single SG, or define an SG per mobile object. In the context of current standards, the latter option is better suited in specific cases, for example when few objects are likely to appear simultaneously in the field of the camera. Indeed the number of SG limited to 8 (even for the extended profile of H.264) implies a maximum of 7 possible moving objects in an image (the last group slices being dedicated to the background). In the case of H.264, this image-by-image update of the slices group is accompanied by the transmission of a parameter better known by the abbreviation PPS (Picture Parameters Set) which indicates to the decoder receiving the stream the new slicing of the image. Thanks to this splitting into slices, the coder carries out a constrained motion estimation for the current group of slices. Thus, the motion estimation vectors will not point to blocks of another group of slices or group slices, allowing two independent reconstructions of the first and the background. The current result is therefore, in this exemplary implementation, a video sequence comprising for each image at least two groups slices, one containing the background and the other all moving objects, with independent predictions between each group slices.

The protection of the accessibility of the information contained in the image can then be achieved by targeting the slices groups comprising the foreground.

The slices group comprising the background will be directly transmitted to a compression step 6. The slice (s) group (s) comprising the moving objects will in turn be compressed at first 8b and then encrypted 8a. It is also possible to associate different security levels with independent mobile objects, and depending on these levels to adapt the encryption step (number of modified bits, different keys, etc.). The two outputs of steps 6 and 8a then form a single compressed and partially encrypted stream Fc to which a NAL of undefined type or undefined type 30 and 31, within which it is possible to transmit any type of information, has been added to indicate the slices that have been encrypted. Unlike other types of NAL, NAL 30 and 31 are not reserved for the flow itself or for RTPRTSP type network protocols. A standard decoder will simply set aside this information whereas a specific decoder, developed to take into account these NAL, may choose to use this information to decrypt the portions (the slices) encrypted. Since the first step made it possible to target mobile objects during encryption, the use of a standard decoder will produce a sequence in which the first plan will be visually encrypted with a normal background, whereas, in the case of a decoder using the encryption or encryption key and the information identifying the portions of the encrypted stream, the images will be fully intelligible, readable. The encryption step can use the technique described in the patent application WO 2006/067172 for modifying the bits in the image that will allow decoding with a standard decoder (without decryption) or with a decoder using the encryption key or encryption. FIG. 6 is a block diagram of an exemplary system according to the invention comprising a video encoder adapted to execute the steps described in FIG. 5.

In the figure is shown only the video transmitter portion 10 for the transmission of a stream of images being compressed. The transmitter comprises a video coder 11 adapted to analyze, according to the steps described in FIG. 5, the different zones belonging to the foreground and other zones belonging to the background of a video image, to define the slice groups for the next image and compress it. The module 13 determines within the stream the slices groups one or more modules 12 will protect the confidentiality level by encryption or encryption before the transmission module 14 does not broadcast.

Without departing from the scope of the invention, other encryption or encryption techniques than those detailed in the present description, for encrypting zones in an image to ensure the confidentiality of the information they contain, can be used .

This visual encryption operation can also be performed in parallel with a target bit rate order of the moving objects in the sequence. It will then be possible to add robustness to errors (which increases the bit rates) coupled to an adaptation to a target bit rate by optimization during compression or on the fly on the already compressed stream. This coupling will for example maintain a fixed rate while protecting the content related to the foreground. The method and the system according to the invention have the following advantages in particular: the fact of using the analysis in the compressed domain makes it possible to determine the areas that a user wishes to protect by visual cryptography. The proposed parameterization of the slices groups makes it possible to respect as closely as possible the shape of the moving objects with a block resolution. This allows video surveillance to transmit the encrypted stream, for example, to a guard post where security officers can identify the presence of a person, a car, a truck, etc. distinctly through a silhouette without being able to directly identify the face of the person or the license plate of the vehicle. This feature allows for wider dissemination of surveillance flows while respecting the constraints of respect for privacy.

Claims (1)

CLAIMS1 - A method for protecting at least part of a video stream or a video sequence at least partially compressed against a violation of the information contained in said stream, said stream being able to be decomposed into a first type of objects and a second type of object, the method applying to each of the images contained in a video sequence, characterized in that it comprises at least the following steps: a) analyzing the video sequence in the compressed domain in order to define for an image data N at least one or more groups of objects to be protected by encryption, designated first group of objects and a second group of objects, (2, 3), b) predict from the data from the analysis to the step a) of the compressed image N, the position of the objects for a next image N + 1, (4a), c) determining the division into portions or groups of portions of the image N + 1, (4b) ), d) compress the parts of the image comprising the group (s) 20 of portions of the image N + 1 linked to the first plane (8b) then encrypt (8a) at least a part of this or these group (s), e) compress (6) the others types of groups of objects for the image N + 1 f) adding (9) to the groups of objects of the image N + 1 from steps 25 d) and e) information identifying the groups of encrypted objects or The method according to claim 1, wherein the step of predicting the position of an object comprises the following steps: using on the one hand the results of the analysis in the compressed domain leading to blocks; identified within an image, and also the stream being compressed, - follow-up from these identified blocks by matching and prediction using the history of the positions of the objects on the N previous images. 3 - Process according to claim 2 characterized in that the pairing step implements a Munkres algorithm. 4 - Process according to claim 2 characterized in that the position prediction step uses a Kalman filter. Process according to one of Claims 1 to 4, characterized in that it implements, for the encryption or encryption step, a selection of the bits in an image allowing decoding with a standard decoder. 6. Process according to one of claims 1 to 5, characterized in that the video sequence being produced by an MPEG-4 standard part 10 / H.264, the method for defining the portions or slices groups uses the scheduling technique. flexible or FMO (Flexible Macrocbloc Ordering) authorizing the definition of groups of portions or slices macroblock groups by macroblock. 7. Process according to claim 6, characterized in that the method associates a group of portions or slices group by object to be protected or moving object. 8. Process according to claim 7, characterized in that it comprises the transmission of a parameter PPS (Picture Parameters Set) which indicates to the decoder the new division of the image for the image-by-image updating of the groups of portions of the image. 9 - Process according to claim 7, characterized in that it comprises the transmission of an undefined NAL-type network transport unit (described in the standard under the name undefined NAL), containing the information indicating whether the slice group has been encrypted or unencrypted for image-by-image updating. System for visually encrypting at least part of an at least partially compressed video stream or video sequence, characterized in that it comprises at least the following elements: a video portion (10) for the transmission of a stream of images being compressed, a video encoder (11), a stream selection module to be encrypted (13) and an encryption module (12) adapted to implement the steps of the method according to one claims 1 to 9, a module (14) for transmitting the stream at least partially encrypted.